The accuracy of computed tomography for the diagnosis of tibial nonunion

Background: When a patient is seen with a possible tibial nonunion and equivocal findings on plain radiographs, the surgeon may choose to obtain a computed tomography scan to better delineate the bone anatomy. However, the sensitivity and specificity of computed tomography in this setting is not known. We investigated the accuracy of computed tomography for detecting nonunion in this clinical situation. Methods: Thirty-five patients with equivocal findings on plain radiographs underwent computed tomography scanning. The patients were first seen at an average of 9.7 months after the injury and had undergone a mean of 2.6 prior operations. A so-called gold standard of union or nonunion was determined by either surgical findings (for twenty-five patients who were operatively treated) or six months of clinical observation (for ten patients who had nonoperative treatment). Computed tomography scans were assessed by two radiologists and one orthopaedic surgeon who were blinded to the clinical outcome. Results: Computed tomography scans displayed very good diagnostic accuracy. Intraobserver agreement was high (intraclass correlation coefficient = 0.89), the sensitivity for detecting nonunion was 100%, and the overall accuracy was 89.9%. Computed tomography was limited by a low specificity of 62%, as three patients who were diagnosed as having tibial nonunion with computed tomography underwent surgery and were found to have a healed fracture. Conclusions: Computed tomography displays very good accuracy in the evaluation of tibial fracture-healing. However, it is limited by low specificity and may sometimes misrepresent a healed fracture as a nonunion. Surgeons must be aware of this pitfall in order to accurately determine which patients need surgical intervention. Level of Evidence: Diagnostic Level I. See Instructions to Authors for a complete description of levels of evidence.